Information

Panel data analysis is concerned with data that run over time for different units of observations. That can be individuals, households, firms, regions, countries, etc. The repeated nature of the data enables the use of more realistic econometric models (for example, incorporating dynamics at the individual level), more robust estimation methods (e.g. controlling for unobserved heterogeneity) and more powerful testing procedures (e.g. testing for long-run purchasing power parity by pooling a number of countries). On the other hand, the panel nature of the data introduces some complications, including possible parameter heterogeneity, serial correlation or error component structures of the error terms, and cross-sectional dependence. These issues are particularly disturbing in nonlinear or dynamic models. The course will provide an insightful and useful elaboration of the state of the art in empirical panel data analysis at an introductory level. Starting from a general introduction, discussing advantages and disadvantages of panel data, static models with fixed effects or random effects, instrumental variables estimators, clustered standard errors and robust inference, the course focuses on (micro-economic) dynamic models, where the time dimension is typically limited and the number of cross-sectional units is large. Topics covered include fixed effects and random effects models, instrumental variables, estimation by the Generalized Method of Moments (GMM), and models with limited dependent variables.